This paper presents a touch-based control method for tracking 3D surfaces in robotic finishing tasks, specifically for the footwear industry. Our method combines a normal force controller and an orientation controller, both based on the forces feedback from a tactile sensor with nine contact points. The controller regulates the contact force and the orientation of a tool mounted on a robotic end-effector by generating velocity commands in Cartesian space, which allows the robot to adapt the tool pose according to the interaction with the surface. The system does not require a predefined position trajectory, instead it generates velocity commands. The results show the controller’s capability to maintain a constant contact force with an error of 0.229 ± 0.169 [N] when testing with a real shoe, and it adapts to unknown surfaces with different curvatures and slopes.

Recent advances in AI have led to significant results in robotic learning, but skills like grasping remain partially solved. Many recent works exploit synthetic grasping datasets to learn to grasp unknown objects. However, those datasets were generated using simple grasp sampling methods using priors. Recently, Quality-Diversity (QD) algorithms have been proven to make grasp sampling significantly more efficient. In this work, we extend QDG-6DoF, a QD framework for generating object-centric grasps, to scale up the production of synthetic grasping datasets. We propose a data augmentation method that combines the transformation of object meshes with transfer learning from previous grasping repertoires. The conducted experiments show that this approach reduces the number of required evaluations per discovered robust grasp by up to 20 %. We used this approach to generate QDGset, a dataset of 6 DoF grasp poses that contains about 3.5 and 4.5 times more grasps and objects, respectively, than the previous state-of-the-art. Our method allows anyone to easily generate data, eventually contributing to a large-scale collaborative dataset of synthetic grasps.

Object handover is an important skill that we use daily when interacting with other humans. To deploy robots in collaborative setting, like houses, being able to receive and handing over objects safely and efficiently becomes a crucial skill. In this work, we demonstrate the use of Reinforcement Learning (RL) for dexterous object handover between two multi-finger hands. Key to this task is the use of a novel reward function based on dual quaternions to minimize the rotation distance, which outperforms other rotation representations such as Euler and rotation matrices. The robustness of the trained policy is experimentally evaluated by testing w.r.t. objects that are not included in the training distribution, and perturbations during the handover process. The results demonstrate that the trained policy successfully perform this task, achieving a total success rate of 94% in the best-case scenario after 100 experiments, thereby showing the robustness of our policy with novel objects. In addition, the best-case performance of the trained policy decreases by only 13.8% when the other robot moves during the handover, proving that our policy is also robust to this type of perturbation, which is common in real-world object handovers. Code and videos can be found here.

The task of grasping objects is a simple and routinely action for humans but it is complex for robots. To integrate robots into everyday tasks, they have to be equipped with capabilities human-like dexterity. In this line, we propose an analytic method, called GeoGraspEvo, to compute grasping points to be used by robotic hands with three, four or more fingers. Our proposal uses features computed from visible surface objects captured by a single RGBD image of a scene. Additionally, it uses as input some configurable kinematic parameters to be able to carry out the grasping depending on the hand morphology. The method compute grasping points with no training process.

Nowadays, corrugated packaging is designed and manufactured to provide protection to products and goods when they are shipped. Packaging machinery is being broadly used by manufacturers in cardboard-based packaging production. The results of this machinery are disassembled packaging on pallets. Monitoring these palletized packaging is also recommended to optimize the logistic distribution inside a factory. In this paper, we present a visual monitoring system to perform visual control of palletized packaging workflow. Our system, based on low-cost sensors and deep learning techniques, facilitates the storage in warehouses. Our proposal detects and recognizes different type of cardboard packaging on pallets. Besides, the system counts the pallets of each type for helping to automatically address them on the roller conveyors. We have tested its behavior on a real scenario at Smurfit Kappa facilities as well as evaluated its precision-performance, which reaches an APIoUof 0.93 at 14 FPS using low cost cameras.

The goal of this paper is to present a robotic workcell to automate several tasks of the cementing process in footwear manufacturing. Our cell's main applications are sole digitization of a wide variety of footwear, glue dispensing and sole grasping from conveyor belts. This cell is made up of a manipulator arm endowed with a gripper, a conveyor belt and a 3D scanner. We have integrated all the elements into a ROS simulation environment facilitating control and communication among them, also providing flexibility to support future extensions. We propose a novel method to grasp soles of different shape, size and material, exploiting the particular characteristics of these objects. Our method relies on object contour extraction using concave hulls. We evaluate it on point clouds of 16 digitized real soles in three different scenarios: concave hull, k-NNs extension and PCA correction. While we have tested this workcell in a simulated environment, the presented system's performance is scheduled to be tested on a real setup at INESCOP facilities in the upcoming months.

In the past, methods for hand sign recognition have been successfully tested in Human Robot Interaction (HRI) using traditional methodologies based on static image features and machine learning. However, the recognition of gestures in video sequences is a problem still open, because current detection methods achieve low scores when the background is undefined or in unstructured scenarios. Deep learning techniques are being applied to approach a solution for this problem in recent years. In this paper, we present a study in which we analyse the performance of a 3DCNN architecture for hand gesture recognition in an unstructured scenario. The system yields a score of 73% in both accuracy and F1. The aim of the work is the implementation of a system for commanding robots with gestures recorded by video in real scenarios.

Tactile sensors provide useful contact data during the interaction with an object which can be used to accurately learn to determine the stability of a grasp. Most of the works in the literature represented tactile readings as plain feature vectors or matrix-like tactile images, using them to train machine learning models. In this work, we explore an alternative way of exploiting tactile information to predict grasp stability by leveraging graph-like representations of tactile data, which preserve the actual spatial arrangement of the sensor's taxels and their locality. In experimentation, we trained a Graph Neural Network to binary classify grasps as stable or slippery ones. To train such network and prove its predictive capabilities for the problem at hand, we captured a novel dataset of ~ 5000 three-fingered grasps across 41 objects for training and 1000 grasps with 10 unknown objects for testing. Our experiments prove that this novel approach can be effectively used to predict grasp stability.

This paper presents a system for the detection of ships and oil spills using side-looking airborne radar (SLAR) images. The proposed method employs a two-stage architecture composed of three pairs of convolutional neural networks (CNNs). Each pair of networks is trained to recognize a single class (ship, oil spill, and coast) by following two steps: a first network performs a coarse detection, and then, a second specialized CNN obtains the precise localization of the pixels belonging to each class. After classification, a postprocessing stage is performed by applying a morphological opening filter in order to eliminate small look-alikes, and removing those oil spills and ships that are surrounded by a minimum amount of coast. Data augmentation is performed to increase the number of samples, owing to the difficulty involved in obtaining a sufficient number of correctly labeled SLAR images. The proposed method is evaluated and compared to a single multiclass CNN architecture and to previous state-of-the-art methods using accuracy, precision, recall, F-measure, and intersection over union. The results show that the proposed method is efficient and competitive, and outperforms the approaches previously used for this task.

This paper presents a description and comparison of two segmentation methods for the oil spill detection in the sea surface. SLAR sensors acquire video sequences from which snapshots are extracted for the detection of oil spills. Both approaches are segmentation based on graph techniques and J-image respectively. Finally, the aim of applying both approaches to SLAR snapshots, as shown, is to detect the largest part of the oil slick and minimize the false detection of the spill.

Intentional oil pollution damages marine ecosystems. Therefore, society and governments require maritime surveillance for early oil spill detection. The fast response in the detection process helps to identify the offenders in the vast majority of cases. Nowadays, it is a human operator whom is trained for carrying out oil spill detection. Operators usually use image processing techniques and data analysis from optical, thermal or radar acquired from aerial vehicles or spatial satellites. The current trend is to automate the oil spill detection process so that this can filter candidate oil spill from an aircraft as a decision support system for human operators. In this work, a robust and automated system for candidate oil spill based on Recurrent Neural Network (RNN) is presented. The aim is to provide a faster identification of the candidate oil spills from SLAR scanned sequences. So far, the majority of the research works about oil spill detection are focused on the classification b etween real oil spills and look-alikes, and they use SAR or optical images but not SLAR. Furthermore, the overall decision is usually taken by an operator in the research works of state-of-art, mainly due to the wide variety of types of look-alikes which cause false positives in the detection process when traditional NN are used. This work provides a RRN-based approach for candidate oil spill detection using SLAR data in contrast with the traditional Multilayer Perceptron Neural Network (MPNN). The system is tested with temporary data acquired from a SLAR sensor mounted on an aircraft. It achieves a success rate in detecting of 97%.

This paper presents a method to analyse 3D planar surfaces and to measure variations on it. The method is oriented to the detection of deformations on the elastic object surfaces formed by flat faces. These deformations are usually caused when two bodies, a solid and another elastic object, come in contact and there are contact pressures among their faces. Our method describes a strategy to model the shape of deformation using a mathematical approach based on two concepts: Histogram and Map of curvature. In particular, we describe the algorithm for deformations in order to use it in visual control and inspection tasks for manipulation processes with robot hands. Several experiments and their results are shown to evaluate the validity and robustness of the method to detect and measure deformations in grasping tasks. To do it, some virtual scenarios were created to simulate contacts with fingers of a hand robot.

Designing educational resources allow students to modify their learning process. In particular, on-line and downloadable educational resources have been successfully used in engineering education the last years [1]. Usually, these resources are free and accessible from web. In addition, they are designed and developed by lecturers and used by their students. But, they are rarely developed by students in order to be used by other students. In this work-in-progress, lecturers and students are working together to implement educational resources, which can be used by students to improve the learning process of computer networks subject in engineering studies. In particular, network topologies to model LAN (Local Area Network) and MAN (Metropolitan Area Network) are virtualized in order to simulate the behavior of the links and nodes when they are interconnected with different physical and logical design.

This paper describes a study and analysis of surface normal-base descriptors for 3D object recognition. Specifically, we evaluate the behaviour of descriptors in the recognition process using virtual models of objects created from CAD software. Later, we test them in real scenes using synthetic objects created with a 3D printer from the virtual models. In both cases, the same virtual models are used on the matching process to find similarity. The difference between both experiments is in the type of views used in the tests. Our analysis evaluates three subjects: the effectiveness of 3D descriptors depending on the viewpoint of camera, the geometry complexity of the model and the runtime used to do the recognition process and the success rate to recognize a view of object among the models saved in the database.

Image Based Visual Servoing (IBVS) is a robotic control scheme based on vision. This scheme uses only the visual information obtained from a camera to guide a robot from any robot pose to a desired one. However, IBVS requires the estimation of different parameters that cannot be obtained directly from the image. These parameters range from the intrinsic camera parameters (which can be obtained from a previous camera calibration), to the measured distance on the optical axis between the camera and visual features, it is the depth. This paper presents a comparative study of the performance of D-IBVS estimating the depth from three different ways using a low cost RGB-D sensor like Kinect. The visual servoing system has been developed over ROS (Robot Operating System), which is a meta-operating system for robots. The experiments prove that the computation of the depth value for each visual feature improves the system performance.

Traditional visual servoing systems have been widely studied in the last years. These systems control the position of the camera attached to the robot end-effector guiding it from any position to the desired one. These controllers can be improved by using the event-based control paradigm. The system proposed in this paper is based on the idea of activating the visual controller only when something significant has occurred in the system (e.g. when any visual feature can be loosen because it is going outside the frame). Different event triggers have been defined in the image space in order to activate or deactivate the visual controller. The tests implemented to validate the proposal have proved that this new scheme avoids visual features to go out of the image whereas the system complexity is reduced considerably. Events can be used in the future to change different parameters of the visual servoing systems.

This paper analyzes the learning experiences and opinions obtained from a group of undergraduate students in their interaction with several on-line multimedia resources included in a free on-line course about Computer Networks. These new educational resources employed are based on the Web2.0 approach such as blogs, videos and virtual labs which have been added in a web-site for distance self-learning.

The depth perception in the objects of a scene can be useful for tracking or applying visual servoing in mobile systems. 3D time-of-flight (ToF) cameras provide range images which give measurements in real time to improve these types of tasks. However, the distance computed from these range images is very changing with the integration time parameter. This paper presents an analysis for the online adaptation of integration time of ToF cameras. This online adaptation is necessary in order to capture the images in the best condition irrespective of the changes of distance (between camera and objects) caused by its movement when the camera is mounted on a robotic arm.

This paper describes a method to detect changes in given surfaces tracked by the robot end-effector. To do so, first an approach to combine visual and force information is described. By filtering the interaction forces, a parameter which provides the probability of a change in the tracking surface is obtained. In order to obtain a more accurate determination of the change in the surface, information from a laser located at the robot end-effector is also employed. Therefore, a robust method for detecting surface discontinuities which employs force and visual information jointly with structured light is presented